So in the first lecture I stressed the importance of protein phosphorylation for biology and in the second lecture I tired to articulate for you the protein kinase molecule, the three-dimensional molecule and how it works. What I’d like to do here is focus on how that kinase is regulated and how it is localized. And so again, I’m going to focus on PKA as the model system in part because it’s the kinase we really understand the most about in biochemical terms and structural terms. So, again, it is activated by a hormone binding to a GPCR, which then leads to the generation of cyclicAMP and then that unleashes the catalytic activity of PKA, the kinase. What I’m going to focus on today is this other side where it is in its inactive state and also targeted to specific sites in the cell by A Kinase Anchoring Proteins. This is really an essential feature of specificity and biological function for PKA. And so we’re going to look at the holoenzymes. I showed you previously what we learned from the structure of the catalytic subunit by itself. What I’m going to tell you here is how much more information we obtained from having a structure of the holoenzyme complex. And then I’m going to focus on the A Kinase Anchoring Proteins and tell you how they target PKA to specific sites and again really leave you with the global challenge for all of us to understand how these kinases exist as part of larger, often very large, macromolecular complexes. So, if we look at that kinase that I told you about before, we have the structure of PKA bound to an inhibitor peptide and I told you about the active site residues and what I want to look at now… Here’s the kinase with the two lobes with ATP bound at the cleft between the two lobes and I’ve told you about those subdomains. I think you have a good concept of those subdomains now. What we’re going to focus on now is the surface of that kinase because it’s the surface that the kinase uses to interact with other proteins. And these kinases are protein interaction modules. They communicate with other proteins and they use their surface to do that. So, here’s the PKA with its conserved subdomains and now if we look at that core of PKA you can see its crevices, the charged features of that surface. Each kinase is quite different on its surface and that dictates what proteins it can interact with. So one of the major proteins that binds to the catalytic subunit are the regulatory subunits and they are multi-functional, modular and highly dynamic. It has two cyclic AMP binding domains at the C terminus and then it has an inhibitor sequence that binds to the active site cleft of the catalytic subunit and then at the N-terminus it has a dimerization domain. This linker region, in the absence of cyclic AMP or catalytic subunit, is a very dynamic entity, this linker region. And so it’s hard to do structure of the whole thing. We could do structures of the monomeric forms and they’re perfectly functional in terms of binding cyclic AMP and binding catalytic subunit with very high affinity, sub-nanomolar affinity. And so we did structures of this one. We’ve also done a structure of the monomeric form. This first structure in 1995 showed us the cyclic AMP binding domain and told us how this regulatory subunit functions as a receptor for cyclic AMP. And cyclic AMP is a major signaling molecule in all of biology. It’s a second messenger signal of external stress in bacteria as well as in man. And the domain that it binds to, this Cyclic Nucleotide Binding domain, is also conserved. So it’s binding to this domain that now translates that cyclic AMP signal into a biological response. So the general features of this domain were really elucidated by this structure of the cyclic AMP bound R subunit and you can see the same domain is conserved in a number of other proteins. So, looking more closely at that motif there is what we call a phosphate binding cassette, this PBC which anchors the phosphate of cyclic AMP–another very important phosphate in biology. And then in addition you have as part of that conserved motif are these B and C helices which are functionally conserved in all of those domains and each cyclic nucleotide binding domain has a hydrophobic capping residue. So you can see that the adenine ring is capped between two hydrophobic residues. And this can come from different part of the molecule but the PBC, the B and C helices are conserved. And so, if we now look at what we have learned from the individual proteins, from cyclic AMP bound structure of the regulatory subunit and from the free catalytic subunit, we’ve learned a lot. How does this function as a receptor? How do we understand that cyclic AMP binding motif? How does it work as a catalyst? So we’ve learned a lot. But we didn’t really understand how is the catalytic subunit actually inhibited by the regulatory subunit nor did we understand how the holoenzyme is activated by cyclic AMP. And these were things that we really have learned just in the last few years and for me it has certainly demonstrated to me the importance of having larger complex to be able to really understand function and certainly we have understood so much more about the function of PKA having a structure of a holoenzyme complex compared to what we knew with just the free catalytic subunits and regulatory subunits. So this is the small cyclic AMP binding domain protein bound to the catalytic subunit and you can see several things from this structure. First, the red part is the inhibitor peptide which was always disordered before. Now it becomes ordered at the active site cleft. The major interactions here of this surface are with the large lobe. So the large lobe is an important docking site for many proteins to bind to protein kinases. OK, so let’s look more closely at what was really surprising to us about this structure. We really didn’t anticipate this at all. So, on the left we have the structure of the cyclic AMP bound domain and then on the right I’m going to show you what happens when you release cyclic AMP and you bind to the catalytic subunit and you can see a major conformational change. The helical subdomain changes and this inhibitor peptide becomes ordered, that is, ordered at the active site cleft. And in this structure that inhibitor region was there but it was always disordered. I told you that linker is very disordered when it’s a free regulatory subunit. So let’s look at first this ordering of the inhibitor site. So I told you in the last lecture how PKI, a small inhibitor peptide, is bound to the active site cleft and they have a common mechanism, this region that’s circled here is the common inhibitor region. It also mimics a substrate. So that binds directly to the active site cleft and they do it in the same way. PKI on the other hand bound with that amphipathic helix. That achieved, again, sub-nanomolar affinity but on a different surface. Whereas here you can see that the regulatory subunit is using a completely different surface to achieve high affinity. And I’m going to show you this again so you can appreciate the dynamic features, especially this B/C helix, how that B/C helix goes from a kinked structure to an extended structure. So, a very, very dynamic module and we didn’t anticipate such a conformational change at all before seeing the structure. So here you can see again that we’ve docked the inhibitor peptide and then you can see as your morph from the cyclic AMP bound conformation into the unbound conformation you can see that C helix, how, as it extends, it binds onto the surface of the large lobe. So, it makes a major contact with the large lobe So this is the activation loop that I told you about before. Here’s that phosphate. This is an active kinase. The regulatory subunit doesn’t bind very well if it’s not an active and phosphorylated kinase. And so, I also told you that this threonine feeds back to all those different parts of the molecule, including the active site. It makes an active kinase. So, if we look where that helix from the regulatory subunit is docking, it’s docking right up against that activation loop and this is a critical interface. So, we now have a further understanding of what you achieve by phosphorylating this kinase. You create an active kinase. So you stabilize the active site so it’s optimized for catalysis. But the other thing that you do based on the holoenzyme structure, is that you also create a surface here that is now providing a docking site for the regulatory subunit. So, again, it impresses, I think on you and all of us, the importance of one single phosphate on the functioning of this protein. This is now both domains and the questions was: does that helix extend in both domains? And of course it does. It’s even longer. And so the consequences of this are really very important in looking at this entire structure. Again, it’s a very large, interface that you have here for the A domain and the B domain binding to the catalytic subunit. And looking at it as a static difference, this is the cyclic AMP bound state where this inhibitor site is present but disordered. You don’t see it. And then, now when you look at releasing cyclic AMP and binding to the catalytic subunit, you now splay apart the two domains. You have a very long extended helix that separates the two domains. And you have the inhibitor peptide docked to the active site cleft. So it’s a major conformational change that you induce in the regulatory subunit as a consequence of releasing cyclic AMP and binding to the catalytic subunit. And the catalytic subunit does not change very much in its conformation. And so this is just letting you see this again and flipping back and forth the change in the proximity and the organization of these two domains. And so you can look, just going back and forth, at how that B domain is really just hooked onto that helix and so when that helix extends, it flips away the B domain and really dissociates the two domains from one another. OK, so we’ve looked at the consequences of binding the regulatory subunit, how the C subunit is inhibited, how it gets activated by cyclic AMP. That came from having a holoenzyme structure. Let’s look at this other feature: the kinase…PKA is not typically just floating around in solution. It’s often targeted to specific sites. And if you want to look at the dynamics of time and space as part of your signaling equation, then you have to look at how is PKA localized. And in large part, this is achieved by A Kinase Anchoring Proteins, scaffold proteins that bind with high affinity to the regulatory subunit. So let me explain that a little bit more. These are just some of examples that John Scott gave me of different places in which the regulatory subunits can be targeted; to the plasma membrane, to mitochondria, to all different sites so that you have localized communities of signaling. And if we look at what I’ve told you so far, it’s this part of the molecule that’s important for binding to the catalytic subunit and inhibiting the catalytic activity. It’s this part, the dimerization domain, that is important for targeting to AKAPs. We’ve looked in particular at Dual Specific AKAPs. Most AKAPs bind to the RII subunits. They bind with very high affinity, in the nanomolar range, to RII subunits. There are some that bind to both RI and RII subunits and those we call Dual Specific AKAPs or D-AKAPs. So, just to again give you a sense of the complexity of this signaling system: This is the L-type calcium channel where you have one of its tails regulated by a PKA phosphorylation site. We’ve had these extraordinary advances in getting structures of channels and membrane proteins and understanding for the first time, at atomic level resolution, how an ion passes through a channel. But, it’s often the tail and the loops that are disordered on the cytoplasmic side that provide the mechanism for regulating, whether that channel is open or not. And this is the case for this L-type calcium channel. There’s a PKA phosphorylation site there and PKA is targeted to this site by this AKAP179 which has many binding partners but it targets to PKA by this little mechanism which is an amphipathic helix in the AKAP. And so you bring the kinase there. You have PP2A as a phosphatase, Calcineurin as a phosphatase. Very often, you have in this system, kinases and phosphatases that work together in a synergistic way. And it’s a synergy which is another thing that I hope to convey to you. How these molecules work together, how different parts of one molecule, like the kinase work together to create an active enzyme, and how different molecules come together to create a signaling unit. So, one example, D-AKAP1, targets PKA to the outer mitochondrial membrane. And if we look at D-AKAP1, these AKAPs have many binding modules and D-AKAP1 has a little PKA binding motif there that I’ll explain to you further. And it also has a targeting region. That’s what takes it to the mitochondria, the N-terminus and if you add an extra 30 residues at the N-terminus it goes to ER. D-AKAP1 is an example of one of these AKAPs. And here you can just see, if you ordinarily have RI in these cells, it’s cytoplasmic. But, if you co-express RI which is labeled with the GFP in green, with D-AKAP1, you find that it all goes to the mitochondria. And this is very characteristic of the mitochondrial morphology in this cell. OK, D-AKAP2 is another one that we’ve looked at. And D-AKAP2 binding both to RI and RII. It regulates ion transporters and so this is another big function of these AKAPs; to regulate ion transporters. And so, the ion transporter that it was first shown to regulate is the sodium-phosphate ion transporter. And this is regulated not by phosphorylation but by internalization. When you don’t want…this is in kidney proximal tubules, when you don’t want phosphate to be taken up, you internalize the transporter. And D-AKAP2 is linked to the transporter though a PDZK domain protein and these are little binding modules (PDZ domains) that now link the AKAP to the co-transporter and PKA, through the AKAP. So, let’s look more closely at D-AKAP2, at just its binding motif. So, the C-terminus of D-AKAP2 contains all the information that important for binding to PKA. It’s this little region in red here. It’s modeled to be an amphipathic helix. It is an amphipathic helix. And then this little bit at the end, the three residues at the end are what target it to that PDZ domain. But it’s going to be this amphipathic helix that we took, this peptide, and crystallized it together with the dimerization domain of RII-alpha. You can see here this complex. The red is the AKAP peptide. It binds with 2nM affinity, very, very tightly. And you can see the N-terminus is a 4 helix bundle of both chains of the regulatory subunit. And then, most recently we did the structure of the RI dimerization domain. Here, where this has a lower affinity, 50nM but critical for docking, particularly under stress conditions, of targeting PKA, the RI subunit. So the RI subunit it typically more soluble than the RII subunits. The RII subunits are mostly targeted. The RI subunits are cytoplasmic but they can be dynamically recruited. If you cap a T cell for example, RI is normally soluble, you can cap it with an antigen. RI goes to the cap site. So, RI is much more dynamic. And it also has disulfide bonds that actually link the two protomers. So, we have a chemical understanding of these docking motifs. And here is just showing you, again, that RI subunit… how this amphipathic helix is binding specifically to this surface and seeing that as a surface, a very close packing of the small helix to this dimerization domain and when you remove the peptide you can see this groove. It’s a very characteristic groove and quite different for RI and RII. So that’s how you achieve targeting. So, let’s look at one more feature. I told you how important it was for us to be able to understand the holoenzyme complex, that complex of a monomeric R subunit and C subunit. So, in the physiological state, RI exists as a dimer. So this is a homodimer of two chains of the regulatory subunit. You have at the N-terminus the dimerization domain that I just told you about. We have the cyclic AMP binding domains at the C-terminus and this very disordered linker region. And so when we look at this linker region you can see first of all, this is all disordered in the free R subunit and you have the inhibitor peptide that binds to the active site cleft. You have other motifs. There are phosphorylation sites, a putative Grb2 binding site, There are many kinds of things that can be going on in this disordered region. So what happens when you bind and make a holoenzyme? Well, we’ve only seen heterodimer: monomeric R with C, but we know under those conditions that this part becomes ordered. So that becomes ordered. So we’re left with a small region that still seems to be quite disordered. So, how do we get an handle on that? So, one way that you can do this is to actually introduce… so this is the part that’s left, you can introduce cysteine residues. Do cysteine scanning mutagenesis and put fluorescent probes in and just look very generally at flexibility. How flexible is that residue when nothing is bound versus when you have a holoenzyme? And when we did this with David Johnson at Riverside, we found that these residues, all of these, in the absence of any ligand, cyclic AMP or C subunit, these were very, very flexible. When you added holoenzymes, some of them stayed flexible but others became much less flexible. So this one at the end, for example, stay flexible. This region here becomes very fixed. It’s part of that tight holoenzyme complex here. But some of these became more fixed in the holoenzyme complex and that led us to think, “Maybe there are some other interactions there.” So, we made a new construct that actually extends further in this N-terminal direction and we crystallized that to see if this communication is going between the two inker regions that we might get some more ordering. And so Angela, a graduate student who carried out this work and got a crystal but it was still a heterodimer. So it had, in one asymmetric unit, a catalytic subunit and a monomeric regulatory subunit but this tail did become ordered and if we looked at that more closely, it didn’t order all of it. There’s still 10 or so residues missing, but it ordered about 10 more residues. And it ordered it in a way that it wasn’t binding to its own cyclic nucleotide binding domain and it wasn’t binding to the catalytic subunit. And so we had to really go to the symmetry related dimer to understand how this might be ordered. And so you can rotate it now. We’re looking at it from the perspective of the bottom of the dimerization domain looking down on this. And so we can see that the N-terminal part of this inhibitor site is ordered by its interaction with the C and B domain of the symmetry related dimer. And, again, this is the inhibitor site for this dimer and the N-terminus of that is coming over and interacting with the C and B domain. And so this provides an incredible opportunity for allostery where we know the Hill coefficient for activating the heterodimers is about 1.2 whereas the Hill coefficient for activating this tetrameric holoenzyme is 2.3-2.4. So there’s an incredible about of allostery that you achieve by having the tetrameric holoenzyme. And one can begin to see how you might achieve that. Here you have the inhibitor peptide docking to both of the symmetry related dimers and then cyclic AMP binding. When it binds to this domain it’s going to unleash that inhibitory peptide from its own bound catalytic subunit. But you can see how it also has the potential to have an allosteric effect on releasing the inhibitor peptide that’s bound to the other heterodimer. So this is very, very exciting for us and although we obviously need to do the structure of the full length holoenzyme and validate this, this structure is of the tetrameric holoenzyme, the model that we create from this is very consistent with much of the information that we have about the tetrameric holoenzyme. So you can see here, as you rotate around, you pick up now a symmetry related dimer which gives you the framework for the tetrameric holoenzyme. As you go around the next time, you pick up the B domain which is highly cooperative for this RI-alpha subunit. You must bind cyclic AMP first to the B domain, then to the A domain and then you release the inhibitor peptide from the active site of the catalytic subunit. And then it’s also a really nice model because the N-terminal part here is positioned so that it can beautifully link up to the dimerization domain. We have only about 20 residues that are missing from this structure now. So it’s a very good model for really beginning to think about a tetrameric holoenzyme and I think as we move forward, one of the real challenges for structural biologists who are working on these signaling proteins is to try to elucidate larger complexes and I think from that we will really gain a much better understanding of how this works as an integrated system. And just an example of the RI subunit that has a class 2 PDZ motif at the C-terminus, which is quite exposed here and we were pretty sure there is some PDZ domain protein that’s going to be binding there as well. Building this higher assembly of molecules is going to be a really important thing for all of us to try to achieve.